>>Sounds basically sound - dbc copied to local is usually faster due to no locking on server,
>>but inside server data dictionary eliminates many pathing hassles.
>>Effect is larger if dbfs are routinely closed again.

The data is not copied local, it is still on the server, test folder is next to live data.

>>Might be oplocking not disabled: caches the single user PC somewhat.
how do I disable oplocking? SMB2 is disabled is that the same?


>>If the path to the data folder is last / after nonexisting paths,
>>loading from there might be slower compared to loading test from the first entry in the path list.
>>Compare times on the normal data path when no other user is logged in
>>Again, effect is larger if dbfs are routinely closed again.
Think I understand - you are talking about my path statement? I use a SET PATH statement, both live and test have same SET PATH except for the data folder.

>>Fragmented real data might be slower than unfragmented test data created by copy.
Live data files all packed 2 days ago, so should not be fragmented.

A standard query on live system, retrieving a dozen or so records out of 30,000-100,000 is taking over 4 seconds on "live" and less than .5 second on "test" data. There are 10+ queries to build a data object so on live it takes 1min+ to load, on test it is taking less than 10 seconds.